OpenAI launches Dots, always-on agents with their own computer
Announced at DevDay, Dots work in the background on GPT-6 Astra, suggest tasks before you ask and live in ChatGPT, Slack and Teams.
OpenAI introduced Dots at DevDay on 29 September 2026. Each Dot is a persistent agent running on GPT-6 Astra with its own cloud computer: a browser, a terminal and a file system. It keeps working when your laptop or phone is off, and connects to your own computer only if you allow it.
You talk to a Dot through group chats in ChatGPT, Slack and Microsoft Teams, and by voice. iMessage and RCS support is on a waitlist. A Dot can use more than 4,000 apps from OpenAI's catalogue, including Gmail, Google Docs, Slack, GitHub and QuickBooks.
The headline feature is proactive research. Without being asked, a Dot looks for work it can take off your plate, such as noticing an unpaid invoice and drafting a reminder for your approval. In this mode it has read-only access and cannot send or change anything.
Control comes from built-in policies, your own Custom Rules and a separate automatic review that watches for actions affecting customers or sharing data. Passwords are stored in a vault the Dot can use but not read, and an activity view shows everything it did.
Dots are available on all ChatGPT Pro plans and on the new Business Premium plan, with Enterprise, Edu and Healthcare in beta on request. Pro is not offered in the European Economic Area, Switzerland or the UK.
What it means for you
If you already pay for ChatGPT Pro outside Europe, you can try a Dot today. Start with read-only tasks and write your rules first.